Interpretable Graduate Employment Competitiveness Classification Using Catboost And Mutual Information Feature Selection
DOI:
https://doi.org/10.24269/mtkind.v20i2.14682Keywords:
CatBoost, Graduate Employment Competitiveness, Mutual Information, Feature Selection, Class Imbalance, Interpretability, Machine LearningAbstract
Graduate employment competitiveness is commonly predicted from academic, skill, and experiential attributes; however, severe class imbalance and data leakage may lead to misleading model performance. This study proposes an interpretable multiclass classification framework that integrates Mutual Information (MI) feature selection with CatBoost, strict data-leakage control, and imbalance-aware evaluation. The experiment used the College Graduates Employment Competitiveness dataset containing 2,000 observations classified into Low, Medium, and High employment-competitiveness classes. The identifier Student_No and the target-related Employment_Competitiveness score were excluded before model development to prevent non-substantive information and target leakage from influencing prediction. MI ranking was performed within the training process, and Top-5, Top-7, Top-9, and Top-11 feature subsets were evaluated using stratified five-fold cross-validation. CatBoost was optimized using class-balanced learning and compared with RBF-SVM and KNN. Model performance was assessed using accuracy, Macro-F1, balanced accuracy, macro precision, macro recall, and multiclass ROC-AUC, while PredictionValuesChange and permutation importance were used for interpretation. The selected Top-7 CatBoost model achieved 97.00% accuracy, 0.8798 Macro-F1, 0.8981 balanced accuracy, and 0.9936 macro ROC-AUC on the independent test set. Industry exposure was the most consistent predictive feature across MI, CatBoost importance, and permutation importance. The results demonstrate that combining MI-based feature selection, strict leakage control, and class-balanced evaluation can provide accurate and interpretable graduate employment-competitiveness classification. Nevertheless, the Low class contained only 60 observations, including 12 test cases, making its performance estimate less stable and indicating the need for validation on larger and more balanced datasets.
Downloads
References
[1] C. L. Caballero et al., “Untangling graduate employability and work readiness to inform evidence-based practice in higher education,” Higher Educ. Res. Dev., vol. 45, no. 5, pp. 1247–1264, 2026, doi: 10.1080/07294360.2026.2615303.
[2] S. Siddique, A. Ahsan, N. Azizi, and O. Haass, “Students’ workplace readiness: Assessment and skill-building for graduate employability,” Sustainability, vol. 14, no. 3, Art. no. 1749, 2022, doi: 10.3390/su14031749.
[3] G. ElSharkawy, Y. Helmy, and E. Yehia, “Employability prediction of information technology graduates using machine learning algorithms,” Int. J. Adv. Comput. Sci. Appl., vol. 13, no. 10, pp. 359–367, 2022, doi: 10.14569/IJACSA.2022.0131043.
[4] R. Haque, A. Quek, C.-Y. Ting, H.-N. Goh, and M. R. Hasan, “Classification techniques using machine learning for graduate student employability predictions,” Int. J. Adv. Sci. Eng. Inf. Technol., vol. 14, no. 1, pp. 45–56, 2024, doi: 10.18517/ijaseit.14.1.19549.
[5] B. S. Andini and K. M. Lhaksmana, “Prediction of student work readiness using artificial neural network and decision tree,” Building Informatics, Technol. Sci. (BITS), vol. 7, no. 1, pp. 222–229, 2025, doi: 10.47065/bits.v7i1.7316.
[6] F. Qi and M. Liu, “An AI-driven reinforcement learning framework for graduate employability prediction and career decision support,” Discover Artif. Intell., vol. 6, Art. no. 911, 2026, doi: 10.1007/s44163-026-01612-6.
[7] H. Zheng, J. Wu, and Y. Zhang, “An explainable hybrid deep learning framework for multi-source student employment prediction,” Sci. Rep., vol. 16, Art. no. 28560, 2026, doi: 10.1038/s41598-026-59158-7.
[8] K. Robindro, U. B. Clinton, N. Hoque, and D. K. Bhattacharyya, “JoMIC: A joint MI-based filter feature selection method,” J. Comput. Math. Data Sci., vol. 6, Art. no. 100075, 2023, doi: 10.1016/j.jcmds.2023.100075.
[9] M. A. Salam, S. Basu Roy, and G. Das, “Efficient approximate top-k mutual information based feature selection,” J. Intell. Inf. Syst., vol. 61, no. 1, pp. 191–223, 2023, doi: 10.1007/s10844-022-00750-4.
[10] J. T. Hancock and T. M. Khoshgoftaar, “CatBoost for big data: An interdisciplinary review,” J. Big Data, vol. 7, Art. no. 94, 2020, doi: 10.1186/s40537-020-00369-8.
[11] M. J. D. Ebinezer and B. C. Krishna, “Explainable artificial intelligence models using SHAP enhanced CatBoost, Bi-GRU with attention, and Tab Transformer,” Sci. Rep., vol. 16, Art. no. 20357, 2026, doi: 10.1038/s41598-026-49961-7.
[12] colabsss, “College Graduates Employment Competitiveness,” Kaggle, [Online]. Available: https://www.kaggle.com/datasets/colabsss/college-graduates-employment-competitiveness. [Accessed: Sep. 19, 2026]
[13] M. A. Salam, S. Basu Roy, and G. Das, “Efficient approximate top-k mutual information based feature selection,” J. Intell. Inf. Syst., vol. 61, no. 1, pp. 191–223, 2023, doi: 10.1007/s10844-022-00750-4.
[14] H. Gong, Y. Li, J. Zhang, B. Zhang, and X. Wang, “A new filter feature selection algorithm for classification task by ensembling Pearson correlation coefficient and mutual information,” Eng. Appl. Artif. Intell., vol. 131, Art. no. 107865, 2024, doi: 10.1016/j.engappai.2024.107865.
[15] J. T. Hancock and T. M. Khoshgoftaar, “CatBoost for big data: An interdisciplinary review,” J. Big Data, vol. 7, Art. no. 94, 2020, doi: 10.1186/s40537-020-00369-8.
[16] Z. Fan, J. Gou, and S. Weng, “A feature importance-based multi-layer CatBoost for student performance prediction,” IEEE Trans. Knowl. Data Eng., vol. 36, no. 11, pp. 5495–5507, 2024, doi: 10.1109/TKDE.2024.3393472.
[17] M. J. D. Ebinezer and B. C. Krishna, “Explainable artificial intelligence models using SHAP enhanced CatBoost, Bi-GRU with attention, and Tab Transformer,” Sci. Rep., vol. 16, Art. no. 20357, 2026, doi: 10.1038/s41598-026-49961-7.
[18] W. Ahmed, M. A. Wani, P. Plawiak, S. Meshoul, A. Mahmoud, and M. Hammad, “Machine learning-based academic performance prediction with explainability for enhanced decision-making in educational institutions,” Sci. Rep., vol. 15, Art. no. 26879, 2025, doi: 10.1038/s41598-025-12353-4.
[19] colabsss, “College Graduates Employment Competitiveness,” Kaggle, [Online]. Available: https://www.kaggle.com/datasets/colabsss/college-graduates-employment-competitiveness. [Accessed: Sep. 19, 2026].
[20] S. Bates, T. Hastie, and R. Tibshirani, “Cross-validation: What does it estimate and how well does it do it?,” J. Am. Stat. Assoc., vol. 119, no. 546, pp. 1434–1445, 2024, doi: 10.1080/01621459.2023.2197686.
Downloads
Published
How to Cite
Issue
Section
Citation Check
License
Copyright (c) 2026 Multitek Indonesia : Jurnal Ilmiah
This work is licensed under a Creative Commons Attribution 4.0 International License.
MULTITEK is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License
